LoRa focusing on Natalie Portman's face, trained on 45 images. Tested with models like realisticvision_v20, dreamshaper, cyberrealistic, reVanimated.

v1.1:Retrained on 1024x1024 pixels. Suitable for generating images with higher resolutions. Best for resolutions between 768 and 1024 pixels.

You can try and mix both loras together to produce better images in lower resolutions or upscale.